REVIEW
Advanced ICD Troubleshooting: Part II
CHARLES D. SWERDLOW* and PAUL A. FRIEDMAN†
From the *Cedars-Sinai Medical Center, Los Angeles, California, and †Mayo Clinic, Rochester, Minnesota, USA
Failure to Deliver Therapy or Delay of Therapy
Absent or delayed therapy may be caused
by programmed values (including human error),
ICD system performance, or a combination of the
two. The most common causes are ICD inactiva-
tion, VT slower than the programmed detection in-
terval, SVT-VT discriminators, undersensing, and
intradevice software interactions. Pacemaker-ICD
interactions, once important, are now rare due to
incorporation of dual- and triple-chamber pace-
makers in ICDs.
ICD Inactivation
If detection is programmed OFF for surgery
using electrocautery, reprogramming must be per-
formed at the end of the procedure, a fact easily for-
gotten, especially in outpatient surgery. One study
reported an unexplained 11% annual incidence
of transient suspension of detection.1 Medtronic
er
ICDs sound an audible Patient AlertTM 6 hours af-
ter the ICD is programmed “off” to alert the patient
to the possibility of inadvertent.2
ac
VT Slower Than the Programmed Detection
Interval
In most ICD patients, VT with cycle lengths
>400–450 ms are tolerated well, while repeated
inappropriate therapies are not. However, slow VT
.p
may be catastrophic in patients with severe LV dys-
function or ischemia.2 All SVT-VT discrimination
algorithms (except ELAs)3–5 deliver less inappro-
priate therapies if the VT detection interval is pro-
w
grammed to a shorter cycle length so that fewer
SVTs are evaluated. To prevent underdetection
of irregular VT, the VT detection interval should
w
be set at least 40–50 ms longer than the slow-
est predicted VT for consecutive-interval count-
ing and 30–40 ms longer for X-of-Y or interval +
w
interval-average counting.5 It should be a longer
cycle length if rapidly conducted SVT is unlikely
or SVT-VT discrimination is reliable at long cycle
lengths (ELA ICDs). A long VT detection interval is
important in patients with advanced heart failure,
Address for reprints: Charles D. Swerdlow, M.D., 8635 W. Third
Street, Ste 1190 W, Los Angeles, CA 90048. Fax: (310) 659-8387;
e-mail: swerdlow@ucla.edu
Received February 15, 2005; revised June 1, 2005; accepted
June 6, 2005.

C 2006, The Authors. Journal compilation 
C 2006, Blackwell Publishing, Inc.
70 January 2006
in whom slow VT may be catastrophic2 (see figure
8 in Part I of this review). The VT detection interval
should be increased if antiarrhythmic drug ther-
m
apy is initiated, particularly with amiodarone or a
sodium-channel blocking (Type 1) drug.6,7 It may
be prudent to measure the cycle length of induced
VT at electrophysiological testing after initiation
co
of drug therapy.6 However, spontaneous VT often
is slower than induced VT.8
SVT-VT Discriminators
SVT-VT discriminators may prevent or delay
d.
therapy if they misclassify VT or VF as SVT.9–12
Discriminators that reevaluate the rhythm diagno-
sis during an ongoing tachycardia, such as stability
and most dual-chamber algorithms, reduce the risk
ic
of underdetection of VT compared with discrimi-
nators that withhold therapy if the rhythm is not
classified correctly by the initial evaluation, such
as onset or chamber of origin. The minimum cycle
length for SVT-VT discrimination should be set to
prevent clinically significant delay in detection of
hemodynamically unstable VT.
Single-Chamber Discriminators
Although spontaneous VT often begins with
irregular R-R intervals, stability algorithms clas-
sify it as VT as soon as R-R intervals regularize,
even transiently. Thus, they rarely prevent detec-
tion of VT (∼0.5%) if they are programmed to nom-
inal settings and the patient is not receiving an
antiarrhythmic drug that decreases cycle length
regularity in VT9,13 (see “Antiarrhythmic Drugs”).
Previously, we noted that single-chamber stability
may be programmed in Medtronic dual-chamber
ICDs to reject atrial fibrillation during redetection.
In this case, stability is applied before the dual-
chamber algorithm. Because stability responds to
irregular rhythms by resetting the VT counter to
zero, the dual-chamber algorithm does not eval-
uate rhythms rejected by stability. Thus, stability
may delay detection of VT even if the ventricular
rate is greater than the atrial rate.
Morphology discriminators also reevaluate
rhythm diagnosis during ongoing tachycardias.
But if they misclassify monomorphic VT initially,
the error usually persists and prevents detection
for the duration of the tachycardia. The St. Jude
morphology algorithm, which analyzes only the
rate-sensing electrode, continuously misclassifies
PACE, Vol. 29
 ADVANCED ICD TROUBLESHOOTING
5–10% of monomorphic VTs as SVT. But, when
it is restricted to tachycardias with ventricular
rate ≤ atrial rate in dual-chamber ICDs, only 2%
of VTs are misclassified.11 The Medtronic mor-
phology algorithm, which usually analyzes high-
voltage electrograms, misclassifies 1–2% of VTs as
SVTs.14 In contrast, the onset discriminator uni-
formly misclassifies VT if it either accelerates grad-
ually across the sinus-VT zone boundary or occurs
during SVT with cycle length in the VT zone.
Dual-Chamber Discriminators
If the atrial lead dislodges to the ventricle, the
atrial channel records a ventricular electrogram.
Any VT will appear to the ICD as a tachycardia
with 1:1 AV relationship and nearly simultaneous
atrial and ventricular activation. Some discrimina-
tors designed to withhold therapy from SVT with
1:1 AV conduction may then withhold appropri-
ate VT therapy. These include the Medtronic 1:1
d.
SVT rule and the St. Jude Rate Branch Algorithm
without additional discriminators, which should
not be programmed until the atrial lead is stable.
VT with 1:1 VA conduction is a difficult diag-
nosis for dual-chamber algorithms. Guidant’s older
Atrial ViewTM misclassifies it as SVT unless the on-
set of VT is abrupt. Medtronic’s PR LogicTM mis-
ic
er
classifies it as SVT in the rare cases in which the
VA interval is both long and constant. Guidant’s
Rhythm IDTM and St. Jude’s BranchTM misclassify
it if a morphology match occurs.11,15
ac
Duration-Based “Safety-Net” Features to Override
Discriminators
These programmable features deliver therapy
if an arrhythmia satisfies the ventricular rate cri-
.p
terion for a sufficiently long duration even if
discriminators indicate SVT (Guidant Sustained-
Duration OverrideTM (SRD), Medtronic High Rate
TimeoutTM , and St. Jude Maximum Time to
w
DiagnosisTM (MTD)). The premise is that VT will
continue to satisfy the rate criterion for the pro-
grammed duration while the ventricular rate dur-
w
ing transient sinus tachycardia or atrial fibrillation
will decrease below the VT rate boundary. The lim-
itation is delivery of inappropriate therapy when
w
SVT exceeds the programmed duration, which oc-
curs in ∼10% of SVTs at 1 minute and 3% of SVTs
at 3 minutes.12,14 The decision to use a discrimi-
nator override should be based on clinical factors
including the probability that discriminators will
prevent detection of VT, the likely consequences of
failure to detect VT, and the likelihood of sustained
SVT in the VT rate zone. For example, override fea-
tures should be considered whenever a morphol-
ogy algorithm is programmed without inducing VT
at electrophysiologic study. The programmed du-
ration in Guidant Atrial ViewTM and St. Jude algo-
PACE, Vol. 29 January 2006
rithms should be increased from the nominal value
of 30 seconds to reduce inappropriate therapies. In
most patients, a duration of 2–5 minutes is reason-
able.
Undersensing
VF may be undersensed due to combinations
of programming (sensitivity, rate, or duration),
low-amplitude electrograms, rapidly varying elec-
trogram amplitude, drug effects, or post-shock tis-
sue changes. Clinically significant undersensing of
m
VF is rare in modern ICD systems if the baseline R-
wave amplitude is ≥5–7 mV.16 Post-shock under-
sensing was an important clinical problem in older
ICD systems using integrated-bipolar leads with
co
closely spaced electrodes.17 Currently, the most
common causes of VF undersensing are drug or hy-
perkalemic effects that slow VF into the VT zone,
ischemia, and rapidly varying electrogram ampli-
tude.4,18 ICDs that adjust dynamic range based on
the amplitude of the sensed R-wave (Guidant) may
be most vulnerable to the latter, extremely rare
problem.18 Prolonged ischemia from sustained VT
slower than the VT detection interval may cause
deterioration of signal quality, resulting in under-
sensing of VF. Lead, connector, or generator prob-
lems may also present as undersensing.
Pacemaker-ICD Interactions
Interactions between ICDs and separate pace-
makers have become rare since ICDs incorpo-
rated dual-chamber bradycardia pacing in the late
1990s. Nevertheless, a few combined systems have
not been revised due to vascular access problems
or other reasons. The multiple potential interac-
tions have been reviewed and testing protocols to
detect them have been developed.19–21 The princi-
pal interaction that may delay or prevent ICD ther-
apy is oversensing of high-amplitude pacemaker
stimulus artifacts. If this occurs during VF, repet-
itive automatic adjustment of sensing threshold
and/or gain may prevent detection of VF.
Intradevice Interactions
Today, “intradevice interactions”—in which
bradycardia pacing features of dual-chamber ICDs
interact with and impair detection of VT or VF—
pose a greater challenge than pacemaker-ICD inter-
actions.22 During high-rate, atrial or dual-chamber
pacing, sensing may be restricted to short periods
of the cardiac cycle because of the combined ef-
fects of ventricular blanking after ventricular pac-
ing and cross-chamber ventricular blanking after
atrial pacing, which is needed to avoid cross talk. If
a sufficient fraction of the cardiac cycle is blanked,
systematic undersensing of VT or VF may occur.
When pacing and blanking events occur at in-
tervals that are multiples of a VT cycle length,
71
 SWERDLOW AND FRIEDMAN

m
co
d.
Figure 1. Failure to detect VT due to an intradevice interaction. The rate-smoothing algorithm in-
troduced atrial and ventricular pacing complexes with associated blanking periods that prevented
detection of VT during post-implant testing. An external rescue shock was required. Shown from
top to bottom are surface ECG, atrial electrogram, ventricular electrogram, and event markers.
ic
At top VT is induced by programmed electrical stimulation with drive cycle length 350 ms and
premature stimuli at 270, 250, and 230 ms (intervals labeled next to event markers). The first
sensed ventricular event occurs 448 ms after the pacing drive (“PVC 448”). The rate smoothing
er
algorithm drives pacing to prevent a pause after the “premature ventricular complex” (PVC),
labeled AP↓1638. A ventricular-paced event does not follow the first AP↓ because a ventricular
event is sensed (VT 415). Subsequent rate smoothing generated atrial and ventricular pacing
pulses (indicated by AP↓ and VP↓ markers, respectively). The resultant post-pacing blanking
ac

periods are shown in the figure as horizontal bars. PAB denotes cross-chamber (post-atrial-pace)
ventricular blanking period. VBP denotes same-chamber (post-ventricular-pace) blanking period.
Together, they prevent approximately 4 of every 6 VT complexes from being sensed. Since the VT
counter must accumulate 8 out of 10 consecutive complexes in the VT zone for detection of VT
to occur, VT is not detected.
.p

ventricular complexes are repeatedly under- VT/VF are complex and difficult to predict, but
sensed, delaying or preventing detection (see they usually elicit a programmer warning. Gen-
Fig. 1).22–24 erally, aggressive rate-smoothing (a small allow-
w

Although uncommon, intradevice interac- able percentage change in R-R intervals), a high
tions have been reported most frequently with the upper pacing rate, a long and fixed AV interval
use of the Rate SmoothingTM algorithm in Guidant favor undersensing and should be avoided. If rate-
w

ICDs.22–24 This algorithm is intended to prevent smoothing is required, parameter settings that re-
VT/VF initiated by sudden changes in ventricular sult in warnings should be avoided by program-
rate.25 It prevents sudden changes in ventricular ming a dynamic AV delay, limiting the upper rate
w

rate by pacing both the atrium and ventricle at in-
tervals based on the preceding (baseline) R-R inter-
val. As an unintended consequence, it may prevent
sensing of VT/VF in some patients because it intro-
duces repetitive post-pace blanking periods. The
algorithm applies rate-smoothing to baseline inter-
vals independent of their cycle length, including
intervals in the VT or VF zones. Intradevice inter-
actions that result in delayed or absent detection
of VT/VF are most common and most danger-
ous when VT is fast. The parameter interrelation-
ships that result in delayed or absent detection of
to less than 125 beats/min, and shortening ventric-
ular blanking after atrial-paced events to maximize
the sensing window for VT/VF. This programming
reduces, but does not eliminate, the risk of under-
sensing.22–24,26
Unsuccessful Therapy
Because defibrillation success is probabilistic,
occasional shocks fail, but failure of more than two
maximum-output shocks is rare if the safety mar-
gin is adequate.27 If an ICD classifies a shock as un-
successful, stored electrograms must be reviewed

72 January 2006 PACE, Vol. 29

 ADVANCED ICD TROUBLESHOOTING
to determine both if the shock was delivered for
true VT/VF and if the shock actually failed to ter-
minate VT/VF. Shocks from chronic ICD systems
that defibrillated reliably at implant may fail to ter-
minate true VT or VF because of patient-related or
ICD system-related reasons. The problem of high-
implant DFTs has been reviewed.3,4 In chronically
implanted systems, most patient-related causes of
unsuccessful shocks can be reversed, but most
system-related causes require operative interven-
tion.
Misclassified Therapy
ICDs misclassify effective therapy as ineffec-
tive if VT/VF recurs before the ICD determines the
VT/VF episode as terminated and reclassifies the
post-therapy rhythm as sinus. Decreasing the du-
ration for redetection of sinus rhythm (St. Jude)
might correct this classification error. However,
this type of misclassification usually does not con-
d.
stitute a clinical problem, while post-shock detec-
tion of nonsustained VT does. An exception oc-
curs when one iteration of antitachycardia pacing
is programmed for the first therapy and a shock for
the second. If antitachycardia pacing terminates ic
VT, but it recurs before the rhythm is classified as
sinus, the recurrent VT will receive a shock in-
er
stead of potentially effective antitachycardia pac-
ing. Programming multiple iterations of antitachy-
cardia pacing can mitigate this problem. However,
the “Smart Mode” in Medtronic ICDs (nominally
ac
“on”) disables antitachycardia pacing if it is classi-
fied as unsuccessful on four consecutive episodes.
Thus, if successful antitachycardia pacing is mis-
classified repeatedly, Smart Mode will disable it.
In this setting, Smart Mode should be programmed
.p
off.
ICDs may also misclassify shocks as ineffec-
tive if the post-shock rhythm is SVT in the VT
rate zone (catecholamine-induced sinus tachycar-
w
dia or shock-induced atrial fibrillation). Solutions
include applying SVT-VT discriminators to rede-
tection of VT, increasing shock strength to prevent
w
shock-induced atrial fibrillation, or administering
β-blockers to slow post-shock SVT.
Patient-Related Factors
w
Factors that raise DFTs reversibly at the cel-
lular level include hyperkalemia, antiarrhythmic
drugs, and ischemia. Pleural or pericardial effu-
sions raise the DFT by forming parallel intratho-
racic current paths that shunt current away from
the heart. Progressive cardiac enlargement or new
myocardial infarction may also raise the DFT and
are less easily reversed. Usually, defibrillation test-
ing should be performed to confirm reliable de-
fibrillation after the suspected cause has been re-
versed.
PACE, Vol. 29 January 2006
A few patient-related causes that would oth-
erwise require operative revision may be resolved
by programming shock pathway and waveform
parameters. For ICDs with fixed-tilt waveforms,
waveform duration depends on output capaci-
tance and pathway resistance.28 ICDs with pro-
grammable waveform duration or tilt (St. Jude)
permit optimization of waveform parameters in-
dependent of pathway resistance. Shortening the
entire waveform might reduce DFT.29 Shortening
phase-two might reduce DFT after amiodarone
m
therapy is started.30 Migration of an active-can
pulse generator low on the chest wall can increase
DFT by altering the shock vector. This may be re-
solved by excluding the pulse generator from the
co
shock circuit, which may be done noninvasively
in Medtronic ICDs.
ICD System-Related Reasons
ICD-related causes include insufficient pro-
grammed shock strength, battery depletion, gen-
erator component failure, lead failure, device-
lead connection failures, and lead dislodgment.
In evaluating data from the ICD interrogation,
attention should be paid to arrhythmia dura-
tion, electrograms from the shocking electrodes,
charge time, battery voltage, the relationship be-
tween programmed and delivered shock strength,
impedance of the high-voltage lead at the time of
shock, and trend plots of lead impedance. Pro-
longed episodes may increase the shock strength
required to convert VT or VF. They may be
caused by delayed detection and/or prolonged
charge times. Programmer software identifies some
system-related problems and displays messages to
alert the clinician.
Problems Identified at Routine Follow-Up
Battery Depletion
Battery Discharge Curves—Relationship to
Elective-Replacement and End-of-Service Indicators
ICDs use batteries composed of lithium an-
odes and silver vanadium oxide (Ag 2 V 4 O 11 ) cath-
odes31 in which reduction at the cathode occurs in
multiple steps. The corresponding discharge char-
acteristics and battery behavior during capacitor
charging have been a source of confusion among
clinicians familiar with the slow, constant voltage
decline typical of lithium-iodine pacemaker bat-
teries. Initially, the cathodal contribution to bat-
tery voltage is due to two simultaneous reactions
corresponding to reduction of V+5 (V+5 + e− →
V+4 ) and Ag+1 (Ag+1 + e− → Ag+0 ). Unloaded cell
voltage falls from about 3.2 to 3.1 V at about 30%
battery depletion. In most ICD batteries, voltage
then falls rapidly to about 2.6 V, where it reaches a
plateau until the battery is 80–90% depleted. This
73
 SWERDLOW AND FRIEDMAN
voltage decrease is caused by a combination of fac-
tors, including build-up of inert material at the
cathode and depletion of the V+5 valence state of
vanadium so that the cathodal half-cell voltage re-
flects reduction of V+4 (V+4 + e− → V+3 ). The 2.6-V
plateau is nominal performance for most (but not
all) ICD batteries. End-of-service indicators corre-
spond to either an unequivocal reduction in un-
loaded voltage about 0.1 V below this plateau or
an excessive increase in charge time.
Premature Battery Depletion
The most common cause is excessive exter-
nal power drain due to pacing or capacitor charg-
ing. Pacing problems include unnecessary ventric-
ular pacing, high pacing outputs, and lead insula-
tion failures. The most common cause of asymp-
tomatic premature battery depletion related to ca-
pacitor charging is repeated aborted shocks due to
repetitive nonsustained VT or oversensing due to
lead-connector problems. Repeated shocks due to
VT storm might also rapidly deplete the battery.
Some diagnostic features have high power con-
sumption. In Medtronic ICDs, “prestorage” of elec-
trograms prior to each VT/VF episode results in
continuous amplification of the unfiltered electro-
gram, substantially reducing longevity. In contrast,
er
prestorage in Guidant and St. Jude ICDs, which
store only filtered electrograms, does not cause sig-
nificant battery depletion. ICD component failure
is a rare cause of battery depletion.
ac
Lead Problems
Ventricular lead problems are the most com-
mon cause of ICD system malfunction.32–35 They
usually present either as oversensing, resulting
.p
in inappropriate shocks, or as abnormal diagnos-
tic measurements at routine follow-up. Presen-
tation as undersensing during VF or unsuccess-
ful therapy is uncommon but often fatal. Clin-
w
ical and lead-design factors identify patients at
high risk for ventricular lead problems and merit
intensified follow-up. Noninvasive assessment is
w
usually diagnostic.36 The introduction of remote
ICD monitoring,37,38 which permits transmission
of complete ICD interrogation via telephone or the
w
internet, may enhance noninvasive identification
of subclinical lead failure (Fig. 2). Patients with
Medtronic ICDs may present with audible “Patient
Alert” warnings (see below).
In ventricular ICDs, atrial lead malfunctions
may present as failure to discriminate VT from
SVT. Atrial lead dislodgment to the ventricle re-
sults in sensing of ventricular electrograms on the
atrial channel, causing detection of one “atrial”
event for each ventricular event. This may cause
dual-chamber algorithms misclassify of VT as SVT
and withhold appropriate therapy. Atrial lead dis-
74 January 2006
lodgment to the ventricle might also result in inap-
propriate shocks caused by ventricular oversens-
ing of noise generated by contact between the leads
(Fig. 3).
Diagnosis of Lead Failure
A systematic approach to the patient with sus-
pected lead failure requires evaluation of elec-
trograms from all electrodes on the lead, pacing
threshold, pacing impedance, and painless high-
voltage electrode impedance; system radiography;
m
stored episode electrograms, data logs, and flash-
back memory (extended recording of R-R intervals
in Medtronic ICDs).36 The yield of these tests is
shown in Table I. The Sensing Integrity CounterTM
co
(Medtronic) displays the number of nonphysio-
logic short intervals detected in the 120–140-ms
range. With intact, true-bipolar leads, the count
remains at zero. With intact, integrated-bipolar
leads, the large proximal sensing electrode (the
d.
distal ICD coil) permits occasional oversensing of
diaphragmatic myopotentials, but a count >300
over 3 months suggests a lead problem.39
ic
Pacing Impedance
The normal value for pacing impedance is a
function of lead design; but for any lead (except
Y-adapted lead combinations), impedance <200
 indicates an insulation defect and impedance
>2,000  suggests a conductor failure if a loose
set screw or faulty adapter are excluded.40,41 Mod-
ern ICDs perform serial impedance measurements
automatically. The range of variability of pacing
lead impedances has been studied for coaxial pace-
maker leads,42 but corresponding data for coaxial
and multi-lumen defibrillation electrodes are lim-
ited. Generally, conductor failures do not present
with moderate changes in impedance. Either the
impedance exceeds 2,000  (continuously or in-
termittently), or it is normal, and conductor failure
is diagnosed by identification of a characteristic
pattern of oversensing. Pacing lead impedance is
insensitive to inner insulation failure of Medtronic
6936 leads.41 Isolated pacing insulation failure of
multi-lumen leads is sufficiently rare that the sen-
sitivity of pacing lead impedance is unknown.
Inner insulation failure in Medtronic 6936 or
6966 leads may be detected by a fall in ring-coil
impedance with normal pacing impedance.41 Re-
view of ring-coil impedance may require sending
the programmer’s “save-to-disk” file to the manu-
facturer for analysis by a specialized software.
High-Voltage Impedance
Shocking-pathway impedance normally is
25–75  in transvenous systems. Values out-
side this range suggest a conductor defect (high
impedance) or insulation failure/short circuit
PACE, Vol. 29
 ADVANCED ICD TROUBLESHOOTING

m
co
d.
ic
er
Figure 2. Asymptomatic malfunction in coaxial defibrillation lead (Medtronic model 6936) detected by internet-based,
remote patient monitoring (Medtronic CareLinkTM ). The Sensing Integrity Counter provides a cumulative count of
ac

nonphysiologic short intervals caused by oversensing. Multiple episodes of nonsustained VT with cycle lengths 140–
210 ms also suggest oversensing. In the absence of exposure to electromagnetic interference, this is highly suggestive of
an inner-insulation failure. The pacing and defibrillation lead impedances are insensitive to early insulation failure.
Analysis of ring-coil impedance from the programmer “save-to-disk” file is often diagnostic. See text. Courtesy of
Physician Assistant Sheila Reynolds and Dr. Dwight Reynolds.
.p

(low impedance). Periodic assessment of the a high-energy shock. If a proximal coil is present,
impedance of the high-voltage electrodes using its integrity is not assessed. Medtronic MarquisTM
w

weak test pulses may detect a loss of lead integrity
before inappropriate or ineffective therapies occur.
In Medtronic and Guidant ICDs, painless pulses
are delivered automatically at periodic intervals
w
ICDs and newer models report independent prox-
imal and distal coil impedances that closely ap-
proximate those of high-voltage shocks. When a
lead defect is suspected, a full-output shock may

to create trend plots. In St. Jude ICDs, high-voltage be necessary to evaluate lead integrity. Partial insu-
lead impedance is assessed by delivering a 12-V lation defects may not be identified by low-energy
test pulse using the programmer command. Pa- pulses that deliver insufficient current to activate
w

tients feel this pulse and may perceive it to be un-
comfortable.
High-voltage impedance measured by weak
pulses correlates well with values measured
by high-energy shocks.43,44 In Guidant ICDs,
impedance values recorded painlessly are compa-
rable to those recorded during high-voltage shock.
In older Medtronic ICDs, the pulse is delivered be-
tween the lead tip and the distal coil, resulting
in nominal low-voltage impedance values of 11–
20 , significantly lower than the impedance of
the shorted high-output protection feature (see
below).
Electrogram Amplitude (R-Wave, P-Wave) and Pacing
Threshold
The amplitude of the R-wave during sinus
rhythm should be at least 5–7 mV for assurance
that the low-amplitude electrograms during VF
will be sensed reliably.16 Causes of reduced elec-
trogram amplitude include antiarrhythmic drugs,
myocardial infarction, shocks, fibrosis at the lead

PACE, Vol. 29 January 2006 75

 SWERDLOW AND FRIEDMAN

m
co
d.
ic
Figure 3. Atrial lead dislodgment causing inappropriate shock. Top right box shows the episode summary, indicating
an episode detected as VF (via the combined count), for which the first VF therapy was delivered. The interval
er
summary plot (middle panel) demonstrates a regular “atrial” rate (boxes) of 640 ms, with a highly variable ventricular
cycle length (black circles), leading to detection and delivery of a 34.6-J shock. The bottom panel shows the atrial
electrogram (Atip to Aring), near-field ventricular electrogram (Vtip to Vring), calculated A-A intervals, event markers,
and calculated V-V intervals. Each ventricular electrogram is associated with an atrial-channel electrogram that has
ac

large amplitude and slope. The sensing of ventricular events on the atrial channel in the absence of sensed atrial events
suggests atrial lead dislodgment to the ventricle. The presence of smaller deflections on the ventricular electrogram
(the first is sensed as “VS 550”) is consistent with oversensing of “noise” signals caused by contract with the atrial
lead. This oversensing results in detection of VF and delivery of shocks during normal rhythm. AR = atrial refractory
sense; VS = ventricular sense; TS = tachycardia sense (in VT zone); VF = fib sense (in VF zone).
.p

tip, new conduction defects, and lead dislodg- resynchronization ICDs dislodge proximally into
ment. These same factors may also elevate the the coronary sinus or right atrium.
w

pacing threshold. The introduction of automatic
capture assessment45 in conjunction with remote
monitoring systems that permit review of device
measurements and trend plots between clinic vis-
w
Subclavian crush occurs when a lead is com-
pressed in the narrowly confined space between
the first rib and clavicle;47,48 this common site of
fracture requires careful examination if lead failure

its37,38 may further facilitate early diagnosis of lead is suspected (Fig. 4). Radiography may also detect
problems. conductor defects, inadequate pin insertion into
the header, abandoned or incompletely removed
w

leads, or torsion due to Twiddler’s syndrome. How-

Radiography
Abnormal radiographic features are identified
in less than half of patients with ICD lead malfunc-
tions.36,46 Lead dislodgment is diagnosed when the
lead tip is in a different position than in the post-
operative radiograph, but variability due to car-
diac and respiratory motion may prevent diagnosis
of minor dislodgments. Typically, RV apical leads
dislodge toward the tricuspid valve; atrial leads
dislodge into the RV; and coronary venous leads of
ever, some intact electrical connections in leads
and adapters appear radiographically to have con-
ductor discontinuity (“pseudo fracture,” Fig. 5).
Fluoroscopy of an unopened lead or adapter (if
available) can be used to determine the expected
radiographic appearance.
Real-Time Telemetry
Real-time display of intracardiac electrograms
is used to assess the effect of body position and

76 January 2006 PACE, Vol. 29

 ADVANCED ICD TROUBLESHOOTING

TABLE I.
Detection of Malfunction in Transvenous ICD Leads

Test Sensitivity*
Asymptomatic Patients with All Patients with
Patients Inappropriate Shocks Lead Failures
(n = 11) (n = 7) (n = 18)

Individual tests

m
Stored electrograms 27% 100% 56%
X-ray 27% 29% 28%
Pacing threshold 36% 14% 28%
R-wave amplitude 55% 0% 6%

co
Pacing impedance 9% 0% 6%
High-voltage lead impedance 29% 14% 43%
Defibrillation threshold 0% 0% 0%
Combination of tests
All pacing/sensing tests 82% 14% 78%

d.
Pacing/sensing, stored electrograms and x-ray 100% 100% 100%
Sensing and detection of induced VF 29% 14% 43%

Modified from Luria et al.36

maneuvers on recorded signals. The appearance of

ic
er
noise during Valsava or inspiration may be caused
by myopotential oversensing49 or a structural lead
defect. Abnormalities of pacing function, R-wave
amplitude, pacing and shocking impedance, and
stored evidence of oversensing39 usually indicate
ac
electrogram (between shocking electrodes), and
sensing/pacing markers may permit localization
of the malfunction to a specific component of the
lead. Medtronic ICDs permit use of special teleme-
try Holter monitors for troubleshooting.50 They

provide a simultaneous, 24-hour recording of a sur-

lead defect. Simultaneous surface ECG recording face ECG lead with two intracardiac electrograms
with telemetry of the sensing signal, the far-field and an extended Marker ChannelTM (Fig. 5).
.p
w
w
w

Figure 4. Subclavian crush injury to defibrillation electrode. (A) Note the sharp bend on the lead
at the site where it passes between the subclavian vein and the first rib (arrow). (B) Lead extracted
from a patient with subclavian crush. Note the disrupted insulation and the disarray of the filers
(the circumferentially coiled conductor) at the site of crush. Reprinted with permission from Lloyd
et al.46

PACE, Vol. 29 January 2006 77

 SWERDLOW AND FRIEDMAN

m
co
d.
ic
er
Figure 5. Oversensing identified by telemetry Holter monitor. Telemetered ECG and ventricular
ac

integrated-bipolar and high-voltage electrograms are recorded with extended marker channel
showing usual markers (top line) and multiple other device parameters. The sixth line shows the
VF counter which increments from 0 to 3 during oversensing. Saturation of the sensing electrogram
and intermittent oversensing are characteristic of electrode problems, in this case intermittent
conductor failure. The Holter was performed because the Short Interval Counter was abnormal.
.p

In this case, the lead impedance was normal.

Stored Episodes shocks delivered per episode, or all therapies in a

zone delivered. Most serious problems identified
w

Lead, adapter, or set screw malfunction often

presents as oversensing, resulting in inappropri- by alerts are lead-related40 (Fig. 6). Unfortunately,
ate detection of VF and aborted or inappropriate ICD recipients may not hear alert tones or may not
shocks. This results from make-break contact noise recognize their significance. Alerts are a useful ad-
w

or oversensing of extracardiac signals due to insu- junct to, but not a substitute for, periodic device
lation defects. The pattern of oversensing often in- interrogation and follow-up.
cludes extremely short nonphysiological intervals
w

(<140 ms, Fig. 6). In Medtronic ICDs, the combi-
nation of a high Sensing Integrity CountTM of non-
physiological intervals, nonsustained tachycardia
episodes with nonphysiological intervals, and ab-
normal lead impedance triggers a lead alert warn-
ing on interrogation.39
Patient Alerts
Medtronic ICDs emit patient-alert tones for
pacing or high-voltage lead impedances out of
range, low battery voltage, long charge time, three
Risk Factors for Lead Failure
Clinical factors predictive of lead failure in-
clude epicardial leads,34,51 abdominal pulse gener-
ator location,36,47 coaxial defibrillation leads,36,41
subclavian venous access,48,52 and dual-chamber
(as opposed to single-chamber) ICD systems.52 In
epicardial systems, sensing and defibrillation are
performed by different leads, eliminating over-
sensing as an early warning for failure of defibrilla-
tion leads. Thus, periodic assessment of shocking
lead impedance is essential. Currently, epicardial

78 January 2006 PACE, Vol. 29

 ADVANCED ICD TROUBLESHOOTING

m
co
d.
ic
Figure 6. Conductor failure identified by lead alert. Top panel shows marked, abrupt increases
in all measured impedance values triggering a lead alert on July 28, 2 days after the patient was
er
seen in clinic. The patient did not hear the alert tones, probably because the ICD was implanted
submuscularly. Middle panel shows episode summaries from inappropriately detected episodes
of VF during intense stationary bicycle exercise in “spinning” class (August 1). Bottom panel
shows the corresponding stored electrogram from the true-bipolar sensing lead. Signals are prob-
ably caused by mechanical motion at the fracture site. Saturation of rate sensing electrogram is
ac

characteristic of conductor failure; but persistence of oversensing throughout the cardiac cycle is
not and is probably caused by intense exercise.

systems are used only in patients with a univen- layer.36,57 A drop in the ring-to-coil impedance
.p

tricular heart or mechanical tricuspid valve. Sub-
cutaneous patches, which have similar construc-
tion to epicardial patches and have similar high
failure rates, also warrant careful observation.34,53
w
is an early marker of this failure,41 which results
in oversensing of nonphysiological signals. Use of
ring-to-coil impedance, the nonsustained episode
log, and Sensing Integrity Counter may permit

Abdominal pulse generator placement, indepen- identification of early failure before oversensing
dent of epicardial or transvenous approach, may causes inappropriate shocks (Fig. 2). A “signa-
cause lead stress associated with tunneling and ture” presentation of this type of lead failure is
w

mechanical friction of the pulse generator against
the lead.36,54 Cardiac resynchronization leads have
a higher dislodgment rate than other electrodes.55
Adapters increase the number of mechanical
w
oversensing of electrical noise following shocks
(Fig. 7).41
Protecting ICDs from High-Voltage, Short Circuits

connections and stress points and may increase
the risk of system malfunction. A loose set screw
(in the header or an adapter) can mimic lead frac-
ture, with make-break contact noise and elevated
impedance.
Coaxial leads have higher failure rates than
multi-lumen leads, in which conductors are longi-
tudinally arrayed along the length of the lead.56
Medtronic TransveneTM coaxial leads (models
6936 and 6966) are prone to failure caused
by metal ion oxidation of a middle insulation
Low impedance on the shock electrodes in-
dicates an insulation failure that diverts shock
current from the heart. It may produce sufficient
peak current that the ICD’s output circuit fails by
“electrical overstress.”58 For example, an insula-
tion failure in the pocket may produce a short cir-
cuit from the high-voltage coil to the can with a
resistance of 8 . A maximum-output (∼800 V)
shock would result in a peak current of 100 A in
the high-voltage output circuit, causing it to fail
catastrophically.59 Shorting between two active

PACE, Vol. 29 January 2006 79

 SWERDLOW AND FRIEDMAN

m
co
d.
ic
er
Figure 7. Post-shock oversensing. Panels show recordings of true-bipolar sensing and marker
channel during VF induced at replacement of an ICD pulse generator attached to a chronically
ac

implanted Medtronic model 6936 TransveneTM , coaxial defibrillation electrode. Top and middle
panels are continuous. At left of top panel, T-wave shock (CD) induces VF, which is reliably sensed,
detected, and terminated by a programmed 20-J shock (CD 19.6 J) in the middle of the middle
panel. Post-shock intermittent oversensing saturates electrogram signals, resulting in failure to
identify sinus rhythm and repetitive inappropriate redetection of VF, with multiple shocks aborted
.p

using the programmer. The lower panel shows one aborted shock with detection of VF (FD). The
short charge time between FD and CE markers is caused by high residual voltage output capacitor
after previous aborted shocks. Post-shock oversensing is the typical early sign of inner insulation
failure in this coaxial defibrillation electrode.
w

intracardiac electrodes of opposite polarity can with no effective means to rescue the patient60 (see
cause the same effect. To prevent electrical- Fig. 8).
w

overstress component failure, modern ICDs protect

the output circuit by aborting the shock pulse im-
mediately if the current in the defibrillation path- Pulse Generator Failure
w

way is excessive. Depending on the manufacturer,
this corresponds to a series lead impedance of 15–
20 . The presumption is that the short circuit
diverts current from the heart, preventing shocks
from terminating VT/VF. In addition to preserv-
ing the generator, this feature has a safety ben-
efit. Simultaneous insulation failure of the rate-
sensing and high-voltage components of RV leads
may present as oversensing. Diverting the resultant
inappropriate shock prevents unnecessary pain.
Further, it may prevent fatal proarrhythmia from
a weak shock delivered in the vulnerable period
Implantable ICD pulse generator failure is
rare, although the incidence is largely unknown
due to limitations in current reporting methods.61
Pulse generator failure might result from primary
semiconductor component failure caused by ex-
posure to extreme voltages or currents (electri-
cal overstress). It can occur due to inappropri-
ate internal arcing, commonly associated with the
hybrid circuit.58 In older ICDs without shorted
high-output protection, pulse generators fail if
shocks are delivered into a short circuit resulting
from lead insulation failure.

80 January 2006 PACE, Vol. 29

 ADVANCED ICD TROUBLESHOOTING

m
co
d.
ic
Figure 8. Protection for shorted output of high-voltage circuit. Text therapy summary indicates sequence of therapies
er
after induction of VF by a 0.6 J T-wave shock (T-shock) during replacement of an ICD pulse generator attached to a
chronically implanted Guidant Model 00125 defibrillation electrode. The measured high-voltage lead impedance is
<20 . Despite programmed shocks strengths of 15, 20, and 30 J, delivered energies are 0.2, 0.3, and 0.3 J, respectively,
because of shorted output protection feature. Short charge times for second and third shocks correspond to high
ac

residual energy on capacitors at end of previous shock. Top left panel shows interval plot. Top right panel shows third
unsuccessful shock (followed by minimum distortion of post-shock, high-voltage electrogram due to low voltage) and
external rescue. An insulation failure in the pocket shorted the high-voltage lead to the generator can.

Pulse generator failure may manifest as ab- periods. These interlocks limit combinations
.p

sence of telemetry, emission of device tones, inap-
propriate shock, premature battery depletion, dis-
play of fault codes upon interrogation, inability to
interrogate or program, failure to charge the capac-
w
of the upper rate limit and AV delay based on
the VT detection interval. Adaptive, rate-related
shortening of the post-pacing ventricular blanking
period permits higher pacing rates with less

itors or retain a charge, early battery depletion, or
failure to deliver therapy. Failed pulse generators
should be explanted and returned to the manu-
facturer for analysis, and the incident should be
w
ventricular blanking, but might increase the risk
of T-wave oversensing.62
In patients with LV dysfunction, RV apical
pacing promotes ventricular dyssynchrony, atrial

reported to the Food and Drug Administration’s fibrillation, and heart failure.63,64 Dual-chamber
Manufacturer and User Facility Device Experience ICDs should be programmed to minimize ventric-
database (http://www.fda.gov/cdrh/maude.html). ular pacing in this population. One approach is
w

Rigorous reporting and analysis may distinguish
random component failure from a systemic failure,
correctable by design or manufacturing modifica-
tions before many patients are affected.61
Considerations for Bradycardia Pacing in ICDs
To prevent intradevice interactions (see
“Pacemaker-ICD Interactions”), dual-chamber
ICDs have interlocks between pacing and VT/VF
detection parameters that limit the fraction of the
cardiac cycle affected by pacing-related blanking
to use a nontracking pacing mode (DDI) and a
long AV interval. However, this may introduce pa-
rameter interlocks or—in conjunction with rate-
smoothing—intradevice interactions resulting in
undersensing of VT/VF. Specific algorithms that
promote intrinsic conduction periodically deter-
mine whether intrinsic conduction is present
during dual-chamber pacing. They either prolong
the AV interval (Search HysteresisTM in Guidant
ICDs; AutoIntrinsic Conduction SearchTM in St.
Jude) or change the pacing mode to a modified

PACE, Vol. 29 January 2006 81

 SWERDLOW AND FRIEDMAN

m
co
d.
Figure 9. Algorithm for maximizing intrinsic AV conduction. (A) Hospital telemetry strip. The
ic
first four complexes demonstrate atrial pacing with intrinsic ventricular conduction in a patient
with MVPTM Mode programmed in a Medtronic ICD. The fifth atrial-paced beat is not conducted,
followed by AV pacing on the sixth beat. This algorithm paces in the AAI mode with ventricular
er
surveillance, to minimize ventricular pacing. Ventricular pacing occurs only if a nonrefractory
sensed or paced atrial event is not conducted (allowing for a maximum pause of two times
the lower rate limit plus 80 ms). Nominal performance of this algorithm may be confused with
pacemaker malfunction. (B) From top to bottom: surface ECG, electrogram markers, and near-field
ac

ventricular electrograms. The first three complexes depict DDDR pacing mode with MVPTM Mode
programmed on. The algorithm periodically checks for the return of AV conduction to minimize
ventricular pacing (seen in the fourth complex, AS-VS). AS = atrial sense; VS = ventricular sense;
VP = ventricular pace.
.p

AAI mode with ongoing assessment of AV conduc-
tion (AAI+ or AAIR+ mode, Medtronic MVPTM ).
The AAIR+ mode eliminates interlocks between
the pacing upper rate limit and VT zone, permit-
w
ing ICDs.65–67 The two widely spaced ventricular
leads in resynchronization ICDs introduce novel
intradevice interactions between resynchroniza-
tion pacing and detection of VT/VF. We address

ting paced rates in the VT zone. In the AAIR+ their unique troubleshooting implications. Opti-
mode, early ventricular events reset the VA inter- mizing pacing for hemodynamic benefit68 is be-
val and ventricular pacing does not occur at a fixed yond the scope of this review.
w

or dynamic AV delay. The AAIR+ mode thus elim-

inates interlocks to permit rate-responsive pac- Ventricular Sensing Problems
ing in the VT zone without the risk of intrade- Early, resynchronization systems utilized
vice interactions that may prevent detection of
w

adapters that have multiple modes of pacing and

VT by competitive bradycardia pacing22–24 (see
“Pacemaker-ICD Interactions”). Because the PR in-
terval is not limited, a single nonconducted P-
wave may occur (see Fig. 9). A potential risk of this
pacing mode is ventricular proarrhythmia caused
by pause-dependent VT.
Troubleshooting Cardiac Resynchronization
Therapy in ICDs
The vast majority of cardiac resynchroniza-
tion pacing in the United States is performed us-
sensing malfunction.68,69 The first dedicated sys-
tem (Guidant Contak CDTM ) sensed between the
LV and RV electrodes to determine R-R inter-
vals. This “extended bipolar” sensing configura-
tion “merges” events sensed in the LV or RV
into a single recording channel. During pacing,
RV and LV depolarizations are synchronized and
refractory periods prolonged, preventing double-
counting. But during conducted rhythms or VT, R-
wave double-counting occurs whenever the inter-
val between local electrograms of conducted beats

82 January 2006 PACE, Vol. 29

 ADVANCED ICD TROUBLESHOOTING

m
co
d.
Figure 10. Multiple shocks in a patient with a cardiac resynchronization ICD with extended
bipolar sensing (Contak CTM ) due to double counting initiated by a pacemaker-mediated tachy-
cardia prevention algorithm. Electrograms from top to bottom are atrial, ventricular near-field,
ic
and ventricular far-field. The initial rhythm is sinus tachycardia, with P-synchronous pacing at
the maximum tracking rate (“VP-MT”). This algorithm periodically extends the post-ventricular
atrial refractory period during upper-rate limit pacing to abort a pacemaker-mediated tachycardia
er
(indicated by “PMT-B”). The following atrial-sensed event falls in the extended post ventricular
atrial refractory period (indicated by the parenthesis, “(AS)”) and is therefore not tracked. Cessa-
tion of ventricular pacing permits double counting of the intrinsic-wide QRS complex by extended
bipolar sensing, leading to inappropriate detection of VF. VS = ventricular sense in sinus zone.
ac

Other abbreviations as in Figure 11.

at the RV and LV electrodes is greater than the ven- promote inappropriate shocks by temporarily sus-
tricular blanking period, leading to double sens- pending ventricular pacing and permitting double-
ing of a single ventricular depolarization.70 Any counting (Fig. 10).
.p

event that inhibits ventricular pacing even tran-
siently and thus permits AV conduction (e.g., pre-
mature complex, conducted sinus tachycardia, or
SVT) may cause R-wave double-counting and in-
w
Extended bipolar ventricular sensing also in-
creases the risk of oversensing P-waves, most
commonly when the LV lead dislodges into the
coronary sinus. This may inhibit pacing, prevent-

appropriate therapy (Fig. 10 [and fig. 4 in Part I of ing resynchronization therapy for heart failure, or
this review]). cause ventricular asystole in patients with com-
R-wave double-counting may result in shocks plete heart block. Oversensing of atrial arrhyth-
w

for SVT slower than the programmed VT detection mias may cause inappropriate shocks despite a
interval because alternate device-detected inter- slow ventricular rate.69 Oversensed events may
vals correspond to (1) the interval between the two have complex and unanticipated interactions with
w

sensed components of the ventricular electrogram detection enhancements (Fig. 11).

and (2) the difference between the true ventricu-
lar cycle length and the double-counting interval.
See Figure 6 for details. Because alternate “R-R”
intervals in any conducted rhythm increment the
VF counter, all detected VT or SVT episodes are
classified as VF and treated with shocks, regard-
less of cycle length. Because there are two detected
intervals for each ventricular electrogram, tran-
sient nonsustained tachycardias may satisfy the
VF detection criterion, resulting in aborted shocks.
Pacemaker algorithms—such as those designed to
terminate pacemaker-mediated tachycardia—may
LV Threshold and RV Anodal Capture
Even if they use RV sensing, most ICDs
use dual-cathodal (“extended bipolar”) pacing in
which the LV and RV tip electrodes serve as the
combined cathode. The common anode may be ei-
ther the RV coil (integrated-bipolar pacing config-
uration) or the RV ring (true-bipolar pacing con-
figuration). Some ICDs permit simultaneous (or
temporally offset) true-bipolar pacing in both the
LV and RV (Fig. 11).

PACE, Vol. 29 January 2006 83

 SWERDLOW AND FRIEDMAN

m
co
d.
ic
er
ac

Figure 11. Extended and true bipolar left LV pacing configurations (available in the Contak Renewal 3TM ICD). The top
two panels depict extended bipolar pacing, in which pacing occurs between LV and RV electrodes. This configuration
has the advantage of requiring only a single LV electrode, but the disadvantage of permitting anodal capture (Fig. 13).
The ability to select either the distal or proximal LV electrode as cathode for pacing functionally permits “moving” the
.p

LV lead via noninvasive reprogramming. This can be advantageous postoperatively if the pacing threshold increases
or phrenic-nerve stimulation occurs. The bottom two panels depict true-bipolar LV pacing, which eliminates the
possibility of anodal RV capture, but requires a bipolar LV lead.
w

Determining whether the LV lead is captur- the pacing pulse may be supra-threshold in one
ing can be difficult. At least seven possible states chamber and subthreshold in the other, leading to
of ventricular pacing can occur in cardiac resyn- loss of resynchronization. In these ICDs, a change
w

chronization systems that use an extended bipolar of ventricular-electrogram morphology during a

pacing configuration. These are no capture (intrin- threshold test indicates loss of capture at one elec-
sic conduction) and capture from six specific pac- trode; the corresponding change in the surface
ing configurations: RV cathodal + LV cathodal (the ECG usually indicates which pacing configura-
w

intended state); LV (cathodal) only; RV cathodal
only; RV anodal only; RV anodal + LV cathodal;
and RV (cathodal + anodal) + LV (“triple site” pac-
ing). Most of these can be distinguished by anal-
ysis of intracardiac electrograms combined with a
12-lead ECG, which is invaluable for follow-up of
resynchronization pacing systems.
In early cardiac resynchronization systems,
dual-cathodal pacing applies the same voltage to
the commonly wired LV and RV cathodes. Since
the local LV and RV thresholds usually differ,
tion is capturing (see below). Current systems per-
mit independent programming of RV and LV out-
puts, facilitating determination of the RV and LV
threshold.
ECG changes associated with biventricular
pacing have been reviewed.68 During determina-
tion of the LV pacing threshold, the QRS complex
usually widens with loss of LV capture. It becomes
less negative in lead I and more negative in lead
III as ventricular excitation originates from the in-
ferior RV as opposed to the lateral LV (Fig. 12).

84 January 2006 PACE, Vol. 29

 ADVANCED ICD TROUBLESHOOTING

m
co
d.
ic
er
Figure 12. QRS morphology during biventricular pacing. The morphology of the paced QRS
complex depends on lead positions and the distribution and properties of diseased myocardium.
ac

Left panel: LV capture is suggested by a markedly negative QRS in lead I. Middle panel: Loss of
LV capture is associated with a positive QRS compels in lead I and a more negative deflection in
lead III. Right panel: With loss of capture, intrinsic conduction returns. In this case, the intrinsic
QRS is narrow.
.p

During dual-cathodal pacing, a ventricular electro- fibrillation coil (integrated-bipolar pacing), which
gram that follows the pacing stimulus by 50–200 has a large surface area (low current density). Thus,
ms may assist in determining the site of capture. It anodal capture is extremely rare in Medtronic and
w

occurs when one ventricular electrode senses the
wavefront propagating from the site of capture in
other ventricle, giving rise to the second event.
“Anodal capture” occurs when the delivered
w
Guidant ICDs, which use integrated-bipolar pac-
ing. St. Jude ICDs use true-bipolar pacing if the LV
lead is unipolar and the RV lead is true-bipolar.
Anodal capture is more common in this pacing

voltage captures at the RV anode either in isola- configuration.

tion or in conjunction with the LV cathode. When
anodal capture results in “triple-site” pacing (LV Ensuring Delivery of Effective Resynchronization
w

cathode, RV cathode, and RV anode), the QRS du-
ration may decrease by 10–20 ms and QRS ampli-
tude of leads I and aVL may increase.71 If anodal
RV capture is mistaken for LV capture, the pac-
ing output may be programmed to a subthreshold
value in the LV, resulting in loss of cardiac resyn-
chronization (Figs. 13 and 14). It is much more
common if the RV anode is a ring electrode (true-
bipolar pacing), which has a small surface area
(high current density) and may be in direct contact
with endocardium than if the RV anode is a de-
In order to provide effective cardiac resyn-
chronization, a high frequency of biventricular
pacing must be delivered. A practical goal is to
resynchronize 90% of R-R intervals. Figure 15
shows a systematic approach to confirming resyn-
chronization. Loss of resynchronization occurs if
LV capture fails. Kay provides a comprehensive re-
view of the differential diagnosis of loss of resyn-
chronization for reasons other than loss of LV
capture.72 The most common causes are intrin-
sic AV conduction due to a long programmed AV

PACE, Vol. 29 January 2006 85

 SWERDLOW AND FRIEDMAN
Figure 13. Possible modes of anodal capture during
biventricular pacing. In all examples, an extended
bipole paces between LV and RV leads. The top left panel
demonstrates biventricular pacing using an integrated-
bipolar RV lead, in which the common anode is the dis-
tal defibrillation coil. Since the RV coil has a large sur-
face area and thus a low current density, anodal capture
is rare. The remaining panels depict biventricular pac-
er
ing between a LV electrode and a true-bipolar RV lead, in
which the common anode is the RV ring electrode. The
top right panel depicts normal function, with capture
only at the LV tip during LV-only pacing. The bottom
ac
left panel depicts capture only at the anode. This oc-
curs if the anodal RV-ring threshold is lower than the LV
threshold and the pacing output is between the RV an-
odal threshold and LV cathodal thresholds. The bottom
right panel depicts anodal and cathodal capture. This
.p
may confound the determination of LV pacing threshold.
See Figure 14.
w
delay and loss of atrial tracking at fast atrial rates.
Others include frequent premature ventricular
complexes, sensing malfunction (see above), pro-
gramming considerations and/or specific pacing
w
algorithms (see below). Interruption of cardiac
resynchronization pacing is common, but it can
usually be restored by noninvasive or invasive in-
w
tervention.73
ICD Programming and Resynchronization Algorithms
Any parameter setting that minimizes ventric-
ular pacing or permits ventricular fusion will ad-
versely affect cardiac resynchronization. Such set-
tings include a long AV delay, a prolonged post-
ventricular atrial refractory period or extension of
the post-ventricular atrial refractory period after
a premature ventricular complex, a low maximal-
tracking rate, AV search hysteresis, rate-smoothing
up or down, or use of a DDI pacing mode. Atrial un-
86 January 2006
dersensing and ventricular oversensing may simi-
larly minimize ventricular pacing and limit resyn-
chronization.
Resynchronization ICDs have features de-
signed to maintain LV stimulation. To prevent LV
T-wave oversensing, Guidant RenewalTM ICDs
incorporate an LV refractory period, which may
inhibit LV pacing. Additionally, they include an
LV protection period after a sensed or paced event
during which pacing will not occur. Although de-
signed to prevent pacing in the LV vulnerable pe-
m
riod, this parameter reduces the maximum LV pac-
ing rate and may inhibit cardiac resynchronization
(Fig. 16).
Resynchronization ICDs also employ algo-
co
rithms designed to maximize ventricular pac-
ing during potentially disruptive events such as
premature ventricular complexes or rapidly con-
ducted atrial arrhythmias. In Medtronic resyn-
chronization ICDs, the Ventricular Sense Res-
d.
ponseTM feature triggers pacing in one or both
ventricles after each RV-sensed event.74 Both
Medtronic and Guidant systems include al-
gorithms that temporarily shorten the post-
ic
ventricular atrial refractory period to regain atrial
tracking and restore resynchronization after pre-
mature ventricular complexes or during sinus
tachycardia faster than the nominal upper rate
limit. Medtronic’s Conducted AF ResponseTM
resynchronizes conducted beats in atrial fibrilla-
tion up to a minimum R-R interval without increas-
ing ventricular rate. Guidant’s Ventricular Rate
RegularizationTM algorithm is intended to restore
resynchronization and ventricular regularity by
pacing the ventricle during irregular conduction
of atrial fibrillation. Concealed conduction into the
AV node from the paced events may slow AV nodal
conduction, thereby limiting the pacing-induced
increase in ventricular rate. These and other algo-
rithms designed to maximize cardiac resynchro-
nization therapy may result in pacing after QRS
onset on surface ECGs and pacing at “unexpected”
times (Fig. 17).
Phrenic-Nerve Stimulation
Left phrenic-nerve stimulation by an LV elec-
trode may make cardiac resynchronization intol-
erable. It may become manifest only after im-
plantation due to postural changes or minor lead
migration. In Guidant ICDs, daily impedance
measurements performed to assess lead integrity
temporarily increase the pacing amplitude. They
may need to be programmed “off” in patients who
have phrenic stimulation at the higher output. De-
creasing the pacing output noninvasively elimi-
nates phrenic stimulation if there is a sufficient
safety margin between the LV and phrenic-nerve
thresholds. Data are insufficient to determine how
PACE, Vol. 29
 ADVANCED ICD TROUBLESHOOTING

m
co
d.
ic
er
Figure 14. Anodal capture during testing of LV capture threshold. LV pacing occurs between a LV cathode and a
RV ring anode. (A) Upper left panel: From top to bottom: surface ECG, electrogram markers and intervals, atrial
electrogram, and ventricular near-field electrogram. At left, pacing output is 2.5 V and anodal capture is present,
with capture occurring at both the LV electrode and the RV ring. No electrograms are visible on the LV channel
ac

due to the simultaneous RV and LV capture. With a decrement in pacing output to 2.25 V, the QRS morphology
changes on the ECG (circled complex); and electrograms appear on the ventricular channel, representing local RV
depolarization. Inset at lower right shows that the time between the pacing stimulus and the RV electrogram represents
the interventricular conduction time during LV pacing. (B) With further decrement in LV pacing output from 0.75 to
.p

0.5 V, true loss of LV capture occurs, with widening of the surface ECG due to left bundle branch block. The local RV
depolarizations are similar during LV pacing (in A) and intrinsic AV conduction (panel B).

often phrenic-nerve stimulation can be eliminated Sensing, Detection, and Pacing

w

by reprogramming from unipolar to bipolar LV pac- Drug effects on the amplitude and slew rate
ing or changing the LV pacing vector on a multi- of local electrograms may impair sensing. This
polar LV lead (Guidant EZTRAK 2†TM ). The alter- may delay or prevent detection of VT/VF. Antiar-
w

native is repositioning the LV lead.
Troubleshooting Interactions with ICDs
w
rhythmic drugs that slow the rate of VT below
the programmed VT rate cutoff can prevent ini-
tial detection of VT2 or divert appropriate ther-
apy during reconfirmation (Fig. 18). Class IC

Antiarrhythmic Drugs drugs bind sodium channels in the open state,

Antiarrhythmic drugs are prescribed com-
monly to ICD patients.75–78 Serious drug-ICD inter-
actions associated predominantly with the use of
sodium- or potassium-channel blocking (Vaughn-
Williams class IA, IC, or III) drugs have been re-
viewed7,76,79 and are summarized in Table II. These
interactions must be considered when antiarrhyth-
mic drug treatment is initiated or the dose is al-
tered.7
resulting in greater drug effect during tachy-
cardias. They may alter ventricular-electrogram
morphology sufficiently in SVT to invalidate
a morphology-algorithm template acquired at a
normal sinus rate. Because VT displays increased
cycle length variability in patients taking Class
IC drugs,9,13 interval-stability detection enhance-
ments may misclassify VT as atrial fibrillation.
The effect of Class IC drugs to increase pacing

PACE, Vol. 29 January 2006 87

 SWERDLOW AND FRIEDMAN

m
co
d.
Figure 15. Systematic approach to insuring delivery of cardiac resynchronization therapy. CRT =
ic
cardiac resynchronization therapy; PVC = premature ventricular complex; VTns = nonsustained
VT; AVN = AV node.
er
ac
.p
w
w
w

Figure 16. Inhibition of LV pacing by the programmable LV protection period in a Contak Re-
newal 3TM . Shown from top to bottom are ECG lead II, RV electrogram, and LV electrogram. A
programmer command for LV pacing was issued during atrial fibrillation. Since the LV protec-
tion period prevented pacing at the selected rate, markers indicating inhibition of pacing appear
(“inh-LVP”). Programming the protection period off permitted assessment of the LV threshold.
RVS = right ventricular sense.

88 January 2006 PACE, Vol. 29

 ADVANCED ICD TROUBLESHOOTING

m
co
d.
ic
er
Figure 17. Algorithms designed to maintain a high percent of cardiac resynchronization pacing during atrial fibrilla-
tion may be confused with inappropriate pacing. (A) Hospital telemetry shows ventricular pacing occurring after the
QRS onset due to the Medtronic Ventricular Sense ResponseTM algorithm, which introduces a triggered biventricular
pacing pulse after each RV-sensed event. (B) Surface ECG lead II, event markers, and the RV tip to LV tip electrogram
ac

are displayed during atrial fibrillation. The first three complexes represent biventricular pacing (“BV”). The last com-
plex represents a conducted complex that was sensed (“VS”). The split marker associated with the “VS” indicates
that a triggered pace pulse resulted. (C) Hospital telemetry shows pacing during rapidly conducted atrial fibrillation
due to an algorithm designed to maximize cardiac resynchronization pacing. This pacing, with the patient at rest, is
distinct from rapid, rate-responsive pacing that is triggered by a sensor.
.p

thresholds is greatest at rapid rates, so that an- Defibrillation

titachycardia pacing could be subthreshold even Antiarrhythmic drugs may cause potentially
if bradycardia pacing is supra-threshold at slower life-threatening rises in DFTs. Drug-defibrillation
w

rates.80 This rarely presents a clinical problem be- interactions are complex. Studies are confounded
cause the pulse amplitude for antitachycardia pac- by anesthetic effects, heterogeneity in ICD leads
ing nominally is higher than that for bradycardia and waveforms, and inter-species variability (e.g.,
w

pacing. When Class IA, IC, or III antiarrhythmic
drugs are started in an ICD patient, the slowest
VT zone detection interval should be increased by
≥40 ms, SVT-VT discriminators should be active
w
human vs canine vs porcine). Drugs that block the
fast inward sodium current (such as lidocaine) el-
evate the DFT, whereas agents those that block re-

polarizing potassium currents (such as sotalol and

to prevent inappropriate therapy of SVT with rate
overlapping that of drug-slowed VT, the interval-
stability discriminator should be programmed to
a less specific value, a new morphology tem-
plate should be acquired during atrial pacing at
a rapid rate, and a sustained-duration override
should be considered. Noninvasive programmed
stimulation may be appropriate to determine if
VT/VF is detected accurately and terminated
promptly.
dofetilide) lower the DFT slightly.7 Improved de-
fibrillation safety margins have mitigated these ad-
verse pharmacologic effects, and potent sodium-
channel blocking drugs are rarely prescribed for
chronic oral therapy in ICD patients. However,
chronic oral amiodarone increase DFTs to a clin-
ically relevant degree in some patients.81 When
treatment is initiated with amiodarone or another
drug that often elevates the DFT, the defibrillation
safety margin should be confirmed if it is both

PACE, Vol. 29 January 2006 89

 SWERDLOW AND FRIEDMAN

TABLE II.
Adverse Interactions Between Antiarrhythmic and ICDs

ICD Function Effect Drugs

Undersensing (rare) Diminished electrogram slew rate/amplitude 1A, 1C

Underdetection of VT VT slowed to rate < detection rate IA, IC, III
Increase in cycle length variability of VT results in
misclassification of VT as SVT by stability enhancement
Inappropriate detection of SVT Organization of atrial fibrillation into atrial flutter with 1:1 1C

m
AV conduction
Use-dependent altered morphology of ventricular 1C
electrogram causes misclassification of SVT as VT

Increase in pacing threshold
Bradyarrhythmias
co
Baseline Amiodarone, 1C
Rapid pacing rates for antitachycardia 1C
Battery depletion from more bradycardia pacing Amiodarone, β-blockers,
calcium antagonists
Pacing-induced dyssynchrony

d.
Increase in defibrillation threshold Unsuccessful shocks/death Amiodarone, 1B
More aborted or delivered shocks Proarrhythmia IA, IC, III

marginal (e.g., safety margin <10 J) and if an in-

sufficient safety margin would result in a change
in therapy (revision of the defibrillation system or
discontinuation of the drug). Similar considera-
ic may contribute to repetitive episodes of VT (“VT
storm”).85
er
tions apply to substantial increases in amiodarone Electromagnetic Interference
dose. Ubiquitous environmental electromagnetic
energy can generate electrical potentials on ICD
sensing electrodes. Such electromagnetic interfer-
Metabolic Effects
ac

ence (EMI) can result in inappropriate detection

Hyperkalemia is the metabolic abnormality
with the greatest clinical effect on ICD func-
tion. In addition to increase in pacing thresh-
old,7,82,83 it can present as T-wave oversensing,83,84
.p
of VT/VF, failure to sense VT/VF, and inappro-
priate pacing or inhibition of pacing. A static
magnetic field causes reversion to magnet mode,
which usually disables detection of VT/VF but

R-wave double-counting, or failure to detect VT or does not alter pacing. Electromagnetic interfer-
VF (Fig. 19). Hypokalemia and hypomagnesemia ence with frequencies less than ∼2 Hz may be
w
w
w

Figure 18. Underdetection of VF caused by antiarrhythmic drugs (amiodarone and mexiletine).

Displayed from top to bottom are the atrial, ventricular near-field and ventricular far-field elec-
trograms, and event markers. VT degenerates to VF. The left panel shows detection of VF. Despite
persistence of VF (seen in far-field ventricular electrograms), slowing in local near-field electro-
gram results in four events at cycle length 378–388 ms, slower than the VT detection interval
(“VS”) resulting in diversion of therapy (“Dvrt Chrg”). Detection occurred eventually, and the
resultant ICD shock terminated VF. Atrial asystole occurs during VT/VF.

90 January 2006 PACE, Vol. 29

 ADVANCED ICD TROUBLESHOOTING

m
co
d.
ic
er
Figure 19. Underdetection of VF caused by hyperkalemia (potassium 6.7) in the setting of chronic
ac

amiodarone therapy. (A) Near-field and far-field ventricular electrograms and a Marker Channel of
a Guidant Prizm VRTM ICD are shown. The top panel shows a sine-wave VT in the far-field channel,
which is detected as VF because local electrograms on the near-field channel are double-counted.
At right of upper panel, four intervals greater than the programmed VF detection interval of 316
.p

ms (190 bpm) result in an aborted shock. Lower panel shows persistence of VF after the shock is
aborted. Too few intervals are sensed in the VF zone to permit detection of VF. The patient was
resuscitated by external shock. (B) ICD system testing after correction of hyperkalemia. Induced
VF is sensed reliably and terminated by an ICD shock. The near-field electrogram is narrow,
w

indicating that the double electrograms present during the clinical arrhythmia were caused by
functional conduction block. Courtesy of Dr. Felix Schnoell.
w

detected as R-waves and prevent automatic ad-
justment of sensitivity from sensing VT/VF. Over-
sensing of higher frequency electromagnetic in-
terference may cause inappropriate detection of
w
Electromagnetic interference is a lesser prob-
lem for true-bipolar sensing than for integrated-
bipolar sensing. It is diagnosed based on history of
exposure and characteristic high-frequency, non-

VT/VF. Pacemaker function may respond as track-
ing (atrial channel) or inhibition (ventricular chan-
nel). These effects have been reviewed.86,87 With
the exception of industrial and military sources,
clinically significant interference by nonmedical
sources is rare.88,89 Interference from sources in
the medical environment has been studied specif-
ically, including magnetic resonance imaging90,91
radiofrequency catheter ablation92 and surgical
electrocautery.93 We address limited issues rele-
vant to troubleshooting.
physiological signals that are larger in far-field
than near-field electrograms (fig. 3 in part I of this
review). Diminishing programmed sensitivity may
mitigate interference. VT/VF detection may be dis-
abled during planned, unavoidable exposure (e.g.,
surgical electrocautery) if an external defibrillator
and trained personnel are present. Field strength
in the patient’s environment may be measured if
external electromagnetic interference is suspected.
Static magnetic fields >5 G should be avoided. ICD
detection of time-varying, external magnetic fields

PACE, Vol. 29 January 2006 91

 SWERDLOW AND FRIEDMAN
depend on their frequency and orientation as well
as the size and number of inductive loops in the
lead. Reasonable field intensity limits are 6000
V/m for 60 Hz alternating current, 70 V/m for
high-frequency fields (>150 kHz), 80 Gauss for
modulated magnetic fields with frequency <10
MHz, and 1 G for those with frequency >10MHz.94
Given the variable response to time-varying fields,
one approach to assessing electromagnetic safety
of an environment is to place an identical and
identically programmed ICD in a saline bath or
conductive gel and expose it to the environment.
The ICD should be connected to the same lead
with several loops to assess the effect of induced
currents. Electrograms and telemetered markers
should be recorded. Alternatively, the patient’s
ICD electrogram and marker channels may be
recorded while the patient is operating equip-
ment, preferably with the ICD in a “monitor-
only” mode and a contingency plan for prompt
defibrillation.
Post-Implant Programming to Optimize Therapy
The medieval jurist Henry de Bracton ap-
plied the phrase “An ounce of prevention is worth
a pound of cure” to philosophy of law. But it
er
is equally applicable to ICDs: Appropriate pro-
gramming minimizes the need for troubleshoot-
ing. In clinical studies, better programming of
SVT-VT discriminators would have prevented ap-
ac
proximately one-quarter of inappropriate therapy
for SVT.11,95 In practice, this fraction probably is
higher. Further, reports continue to include fatal
proarrhythmia and sudden deaths due to ICD pro-
gramming errors.4,60,96
.p
Sensing
If the amplitude of the sinus-rhythm R-wave at
implant is ≥7 mV, nominal sensitivity near 0.3 mV
w
provides adequate sensing of VF.16 More sensitive
settings may be selected if the R-wave is ≤3 mV.
Less sensitive settings decrease the risk of T-wave
w
oversensing in Medtronic ICD, but may increase
the risk of undersensing VF.
Detection Zones and Zone Boundaries
w
Typical values for these rate boundaries are
360–500 ms for slower VT, 340–300 ms for faster
VT, and 280–240 ms for VF. Some experts (includ-
ing the authors) recommend three-zone program-
ming in most patients implanted for secondary
prevention—even those whose only clinical ar-
rhythmia is VF—because most spontaneous VF
begins with rapid VT and most rapid VT can be
terminated by antitachycardia pacing.97,98 Three
zones permit different antitachycardia pacing for
92 January 2006
two distinct rates of VT and antitachycardia pac-
ing for monomorphic VT that overlaps in rate with
polymorphic VT. Other experts recommend two
rate zones with the slowest at cycle length 340–320
ms for patients whose only spontaneous arrhyth-
mia is VF. This approach lowers the risk of inap-
propriate therapy but increases the risk of not treat-
ing VT. In the largest primary-prevention trial,99
ICDs were programmed to a single detection zone
at cycle length 320 ms, without SVT-VT discrim-
inators. Approximately half of shocks were inap-
m
propriate; the incidence of sudden death from un-
detected VT has not been reported.
The sinus-VT rate boundary should be slow
enough to ensure detection of all hemodynami-
co
cally compromising VT. The boundary between
the two VT zones should be based on the cy-
cle length at which different types or fewer tri-
als of antitachycardia pacing are preferred. The
VT-VF rate boundary is based on the cycle length
d.
below which antitachycardia pacing should not
be delivered. In Medtronic ICDs, which use
consecutive-interval counting above the VF Inter-
val and X-of-Y counting below it,62 this bound-
ic
ary should be set to prevent underdetection of ir-
regular, polymorphic VT by consecutive-interval
counting.
Duration for Detection and Redetection
Detection duration prior to antitachycardia
pacing should not be decreased from nominal
values because therapy is immediate after detec-
tion and undersensing of monomorphic VT is
rare. It should be increased in patients who have
long episodes of nonsustained VT. Substantial in-
creases in duration for detection of VT proba-
bly are safe in St. Jude and Guidant ICDs, which
use counting methods that are insensitive to occa-
sional long ventricular intervals or undersensing:
Guidant ICDs use X-of-Y counting, and St. Jude
ICDs count based on the interval and average of the
previous three intervals. In contrast, consecutive-
interval counting used by Medtronic ICDs in the
VT zone may underdetect VT if occasional long
ventricular intervals or undersensing occur.9 Un-
less VT is known to be highly regular, the num-
ber of intervals to detect VT usually should not
be more than 50% greater than the nominal value
in Medtronic ICDs. Duration for detection of VF
may be reduced from nominal only if there is
no alternative method to ensure reliable detec-
tion. Redetection time for VT should be increased
from nominal if VT is well-tolerated, but anti-
tachycardia pacing is not. Redetection time for VF
should be decreased if post-shock undersensing
occurs.
PACE, Vol. 29
 ADVANCED ICD TROUBLESHOOTING
SVT-VT Discrimination
Some SVT discriminators should be pro-
grammed at implant in all patients with intact AV
conduction. See Section III D.
Programmed Therapy
Antitachycardia pacing should be pro-
grammed in all patients unless it is known to
be ineffective or proarrhythmic. See Section IV
C The first shock strength for VF should be set
either in relation to an established implant safety
margin or to maximum.16,27 In modern ICDs with
short charge times, we recommend that shocks for
hemodynamically stable VT be programmed to
the same strength as those for VF, rather than to a
low energy. This maximizes the likelihood that a
ventricular shock will terminate (inappropriately
detected) rapidly conducted, atrial fibrillation. It
also minimizes both the risk that a ventricular
d.
shock will accelerate VT and the risk that it
will be weaker than the atrial upper limit of
vulnerability (and thus initiate atrial fibrillation if
But we recognize that practice differs with regard
to this issue.
ic
it is delivered during the atrial vulnerable period).
er
Evolving Trends
The ICD has expanded from a single-purpose
“shock box” to a multi-purpose platform for elec-
ac
trical cardiac therapies; ICDs under development
include sensors for monitoring heart failure and
other comorbidities. This increasing functionality
and complexity is likely to expand the domain of
troubleshooting and to result in yet unknown, un-
.p
intended consequences, including novel intrade-
vice interactions.
Present ICDs have approximately 500 pro-
grammable features. Despite simplified user in-
w
terfaces that might reduce operator errors, both
nonlethal100,101 and lethal4,96 programming errors
occur. The automatic, self-checking, and self-
w
adjusting functions in today’s ICDs are a wel-
come addition, given the time pressures of present
clinical practice and the fact that troubleshooting
w
is both time consuming and poorly reimbursed.
However, automated features should be disabled
in some conditions. Examples discussed in this
review include the need to disable automatic up-
dating of morphology-algorithm templates in pa-
tients with rate-related aberrancy and the need to
disable Medtronic Smart Mode when VT recurs
rapidly after effective antitachycardia pacing. Fu-
ture automated features may also introduce un-
foreseen consequences that require novel forms of
troubleshooting.
PACE, Vol. 29 January 2006
A potential risk of simplifying the user inter-
face is loss of programming flexibility and trou-
bleshooting tools. ICDs are required to perform
multiple functions based on input signals and clin-
ical conditions that vary widely from patient to
patient. Features optimized for one condition do
not perform optimally for another condition. Con-
sider familiar clinical trade-offs: The preferred so-
lution to rejection of far-field R-waves depends on
the relative sizes and timing of the P-wave and far-
field R-wave as well as the likelihood that the pa-
m
tient will develop atrial fibrillation or flutter. The
preferred solution to discriminating SVT from VT
with a 1:1 AV relationship depends on various fea-
tures such as the reliability of morphology discrim-
co
ination with available electrograms, the difference
between VA and AV times during VT and SVT, and
the specific responses of VT and SVT to antitachy-
cardia pacing. Prophylactic implantation of ICDs
without prior electrophysiological evaluation in-
creases the conflicting requirements for reliable
detection of all VTs while rejecting unanticipated
SVTs.
ICD manufacturers are working to provide a
simplified user interface that provides sufficient
options for common clinical situations. At the
same time, they seek to maintain and increase
the wide range of flexible options that permit op-
timal programming for outlier patients or clini-
cal conditions. Manufacturers must also provide
physicians with adequate technical references for
troubleshooting. Reference manuals are no longer
packaged with ICDs, although some manufacturers
provide online manuals.
Present ICDs provide troubleshooting alerts to
the patient and physician for specific risk condi-
tions.40 Most relate simply to a parameter “out-of-
range,” but at least one lead-integrity alert that in-
tegrates multiple parameters has been verified and
released.39,102 The capability to download ICD data
remotely over the Internet37,38 permits develop-
ment of automated troubleshooting algorithms that
require computing power beyond that available to
implanted ICDs or even ICD programmers. One ap-
plication would be high-level, automated analysis
of stored arrhythmia episodes to evaluate the ac-
curacy of SVT-VT discrimination and recommend
programming changes. Physician-initiated or au-
tomated ICD reprogramming via internet could
both simplify troubleshooting and introduce novel
problems.
Acknowledgments: We wish to thank Jim Gilkerson,
Ph.D. (Guidant), Walter Olson, Ph.D., and Jeff Gillberg, M.S.
(Medtronic), and Tami Shipman, B.S. (St. Jude Medical) for
their diligent reviews of this article.
93
 SWERDLOW AND FRIEDMAN

References 22. Shivkumar K, Feliciano Z, Boyle NG, et al. Intradevice interaction

in a dual chamber implantable cardioverter defibrillator preventing
1. Kolb C, Deisenhofer I, Weyerbrock S, et al. Incidence of anti-
ventricular tachyarrhythmia detection. J Cardiovasc Electrophysiol
tachycardia therapy suspension due to magnet reversion in im-
2000; 11:1285–1288.
plantable cardioverter defibrillators. Pacing Clin Electrophysiol
23. Glikson M, Beeman AL, Luria DM, et al. Impaired detection of ven-
2004; 27:221–223.
tricular tachyarrhythmias by a rate-smoothing algorithm in dual-
2. Bansch D, Castrucci M, Bocker D, et al. Ventricular tachycardias
chamber implantable defibrillators: Intradevice interactions. J Car-
above the initially programmed tachycardia detection interval in
diovasc Electrophysiol 2002; 13:312–318.
patients with implantable cardioverter-defibrillators: Incidence,
24. Cooper JM, Sauer W, Verdino R. Absent ventricular tachycardia de-
prediction and significance. J Am Coll Cardiol 2000; 36:557–565.
tection in a biventricular implantable cardioverter-defibrillator due
3. Shukla HH, Flaker GC, Jayam V, et al. High defibrillation thresholds
to intradevice interaction with a rate smoothing pacing algorithm.
in transvenous biphasic implantable defibrillators: Clinical predic-
Heart Rhythm 2004; 1:728–731.
tors and prognostic implications. Pacing Clin Electrophysiol 2003;
25. Wietholt D, Kuehlkamp V, Meisel E, et al. Prevention of sus-
26:44–48.
tained ventricular tachyarrhythmias in patients with implantable
4. Swerdlow C, Shivkumar K. Implantable cardioverter defibrillators:
cardioverter-defibrillators-the PREVENT study. J Interv Card Elec-

m
Clinical aspects. In: Zipes DP, Jalife J (eds.): Cardiac Electrophysi-
trophysiol 2003; 9:383–389.
ology: From Cell to Bedside, 4th Ed. Philadephia, W.B. Saunders,
26. Shalaby AA. Delayed detection of ventricular tachycardia in a dual
2004, pp. 980–993.
chamber rate adaptive pacing implantable cardioverter defibrilla-
5. Bansch D, Steffgen F, Gronefeld G, et al. The 1 + 1 trial: A prospec-
tor: A case of intradevice interaction. Pacing Clin Electrophysiol
tive trial of a dual- versus a single-chamber implantable defibril-
2004; 27:1164–1166.
lator in patients with slow ventricular tachycardias. Circulation

co
27. Gold MR, Higgins S, Klein R, et al. Efficacy and temporal stability
2004; 110:1022–1029.
of reduced safety margins for ventricular defibrillation: Primary
6. Brode SE, Schwartzman D, Callans DJ, et al. ICD-antiarrhythmic
results from the Low Energy Safety Study (LESS). Circulation 2002;
drug and ICD-pacemaker interactions. J Cardiovasc Electrophysiol
105:2043–2048.
1997; 8:830–842.
28. Block M, Breithardt G. Optimizing defibrillation through improved
7. Goldschlager N, Epstein A, Friedman P, et al. Environmental and
waveforms. Pacing Clin Electrophysiol 1995; 18:536–538.
drug effects on patients with pacemakers and implantable car-
29. Mouchawar G, Kroll M, Val-Mejias JE, et al. ICD waveform op-

d.
dioverter/defibrillators: A practical guide to patient treatment.
timization: A randomized, prospective, pair-sampled multicenter
Arch Intern Med 2001; 161:649–655.
study. Pacing Clin Electrophysiol 2000; 23:1992–1995.
8. Monahan KM, Hadjis T, Hallett N, et al. Relation of induced to
30. Merkely B, Lubinski A, Kiss O, et al. Shortening the second phase
spontaneous ventricular tachycardia from analysis of stored far-
duration of biphasic shocks: Effects of class III antiarrhythmic
field implantable defibrillator electrograms. Am J Cardiol 1999;
drugs on defibrillation efficacy in humans. J Cardiovasc Electro-
83:349–353.
9. Swerdlow CD, Ahern T, Chen PS, et al. Underdetection of ventric-
ular tachycardia by algorithms to enhance specificity in a tiered-
therapy cardioverter-defibrillator. J Am Coll Cardiol 1994; 24:416–
424.
ic physiol 2001; 12:824–827.
31. Takeuchi ES, Quattrini PJ, Greatbatch W. Lithium/silver vanadium
oxide batteries for implantable defibrillators. Pacing Clin Electro-
physiol 1988; 11:2035–2039.
32. Nunain SO, Roelke M, Trouton T, et al. Limitations and late com-
er
10. Weber M, Bocker D, Bansch D, et al. Efficacy and safety of
plications of third-generation automatic cardioverter-defibrillators.
the initial use of stability and onset criteria in implantable car-
Circulation 1995; 91:2204–2213.
dioverter defibrillators. J Cardiovasc Electrophysiol 1999; 10:145–
33. Mattke S, Muller D, Markewitz A, et al. Failures of epicardial and
153.
transvenous leads for implantable cardioverter defibrillators. Am
11. Glickson M, Swerdlow CD, Daoud EG, et al. Optimal combination
Heart J 1995; 130:1040–1044.
of discriminators for differentiating ventricular from supraventric-
ac

34. Glikson M, Luria D, Friedman PA, et al. Are routine arrhythmia

ular tachycardia by dual-chambers defibrillators. J Cardiovasc Elec-
inductions necessary in patients with pectoral implantable car-
trophysiol 2005; 16:732–739.
dioverter defibrillators? J Cardiovasc Electrophysiol 2000; 11:127–
12. Brugada J, Mont L, Figueiredo M, et al. Enhanced detection crite-
135.
ria in implantable defibrillators. J Cardiovasc Electrophysiol 1998;
35. Dorwarth U, Frey B, Dugas M, et al. Transvenous defibrillation
9:261–268.
leads: High incidence of failure during long-term follow-up. J Car-
13. Frang P, Kus T, Vinet A, et al. Underdetection of ventricular tachy-
diovasc Electrophysiol 2003; 14:38–43.
.p

cardia using 40 ms stability criterion: Effect of antiarrhythmic ther-

36. Luria D, Glikson M, Brady PA, et al. Predictors and mode of detec-
apy. Pacing Clin Electrophysiol 1997; 20:2882–2892.
tion of transvenous lead malfunction in implantable defibrillators.
14. Klein G, Manolis A, Viskin S, et al. Clinical performance of
Am J Cardiol 2001; 87:901–904.
waveletTM morphology discrimination algorithm in a worldwide
37. Joseph GK, Wilkoff BL, Dresing T, et al. Remote interrogation and
single chamber ICD population. (abstract) Circulation 2003; 110:III-
monitoring of implantable cardioverter defibrillators. J Interv Card
345.
w

Electrophysiol 2004; 11:161–166.

15. Gold MR, Shorofsky SR, Thompson JA, et al. Advanced rhythm dis-
38. Schoenfeld MH, Compton SJ, Mead RH, et al. Remote monitoring
crimination for implantable cardioverter defibrillators using elec-
of implantable cardioverter defibrillators: A prospective analysis.
trogram vector timing and correlation. J Cardiovasc Electrophysiol
Pacing Clin Electrophysiol 2004; 27:757–763.
2002; 13:1092–1097.
39. Gunderson B, Patel A, Bounds C. Automatic identification of im-
w

16. Swerdlow CD. Implantation of cardioverter defibrillators without

plantable cardioverter-defibrillator lead problems using intracar-
induction of ventricular fibrillation. Circulation 2001; 103:2159–
diac electrograms. Comput Cardiol 2002; 29:121–124.
2164.
40. Becker R, Ruf-Richter J, Senges-Becker JC, et al. Patient alert in im-
17. Panotopoulos P, Krum D, Axtell K, et al. Ventricular fibrillation
plantable cardioverter defibrillators: Toy or tool? J Am Coll Cardiol
sensing and detection by implantable defibrillators: Is one better
w

2004; 44:95–98.
than the others? A prospective, comparative study. J Cardiovasc
41. Ellenbogen KA, Wood MA, Shepard RK, et al. Detection and man-
Electrophysiol 2001; 12:445–452.
agement of an implantable cardioverter defibrillator lead failure:
18. Dekker LR, Schrama TA, Steinmetz FH, et al. Undersensing of VF
Incidence and clinical implications. J Am Coll Cardiol 2003; 41:73–
in a patient with optimal R wave sensing during sinus rhythm.
80.
Pacing Clin Electrophysiol 2004; 27:833–834.
42. Danilovic D, Ohm OJ. Pacing impedance variability in tined steroid
19. Cohen AI, Wish MH, Fletcher RD, et al. The use and interaction of
eluting leads. Pacing Clin Electrophysiol 1998; 21:1356–1363.
permanent pacemakers and the automatic implantable cardioverter
43. Vollmann D, Luethje L, Zenker D, et al. Subthreshold test pulses
defibrillator. Pacing Clin Electrophysiol 1988; 11:704–711.
versus low energy shock delivery to estimate high energy lead
20. Geiger MJ, O’Neill P, Sharma A, et al. Interactions between transve-
impedance in implanted cardioverter defibrillator patients. Pacing
nous nonthoracotomy cardioverter defibrillator systems and per-
Clin Electrophysiol 2003; 26:457–460.
manent transvenous endocardial pacemakers. Pacing Clin Electro-
44. Mehdirad AA, Stohr EC, Love CJ, et al. Implantable defibrillators
physiol 1997; 20:624–630.
impedance measurement using pacing pulses versus shock deliv-
21. Glikson M, Trusty JM, Grice SK, et al. A stepwise testing protocol for
ery with intact and modified high voltage lead system. Pacing Clin
modern implantable cardioverter-defibrillator systems to prevent
Electrophysiol 1999; 22:437–441.
pacemaker-implantable cardioverter-defibrillator interactions. Am
45. Sperzel J, Pitschner HF, Schwarz T, et al. Automatic capture
J Cardiol 1999; 83:360–366.

94 January 2006 PACE, Vol. 29

 ADVANCED ICD TROUBLESHOOTING
verification in ICD lead systems using intracardiac ventricular
evoked response and reduced coupling capacitance. Europace
2003; 5:83–89.
46. Lloyd M, Hayes D, Friedman P. Pacemaker and ICD Radiography.
Cardiac Pacing and Defibrillation: A Clinical Approach. Armonk,
NY, Futura, 2000, pp. 485–519.
47. Kron J, Herre J, Renfroe EG, et al. Lead- and device-related com-
plications in the antiarrhythmics versus implantable defibrillators
trial. Am Heart J 2001; 141:92–98.
48. Gallik DM, Ben-Zur UM, Gross JN, et al. Lead fracture in cephalic
versus subclavian approach with transvenous implantable car-
dioverter defibrillator systems. Pacing Clin Electrophysiol 1996;
19:1089–1094.
49. Babuty D, Fauchier L, Cosnay P. Inappropriate shocks delivered by
implantable cardiac defibrillators during oversensing of activity of
diaphagmatic muscle. Heart 1999; 81:94–96.
50. Feldman CL, Olson WH, Hubbelbank M, et al. Identification of an
implantable defibrillator lead fracture with a new Holter system.
Pacing Clin Electrophysiol 1993; 16:1342–1344.
51. Brady PA, Friedman PA, Trusty JM, et al. High failure rate for an epi-
cardial implantable cardioverter-defibrillator lead: Implications for
long-term follow-up of patients with an implantable cardioverter-
defibrillator. J Am Coll Cardiol 1998; 31:616–622.
52. Kron J. Clinical significance of device-related complications in
clinical trials and implications for future trials: Insights from the
Antiarrhytmics Versus Implantable Defibrillators (AVID) trial. Card
Electrophysiol Rev 2003; 7:473–478.
d.
53. Kuhlkamp V. Initial experience with an implantable cardioverter-
defibrillator incorporating cardiac resynchronization therapy. J Am
Coll Cardiol 2002; 39:790–797.
54. Mehta D, Nayak HM, Singson M, et al. Late complications in pa-
tients with pectoral defibrillator implants with transvenous defib-
55.
ing Clin Electrophysiol 1998; 21:1893–1900.
Daoud EG, Kalbfleisch SJ, Hummel JD, et al. Implantation tech-
niques and chronic lead parameters of biventricular pacing dual-
ic
rillator lead systems: High incidence of insulation breakdown. Pac-
chamber defibrillators. J Cardiovasc Electrophysiol 2002; 13:964–
er
970.
56. Guidant. CardiacRhythm Management Product Performance Re-
port. St. Paul, MN, Guidant Corporation, 2004.
57. Hauser RG, Cannom D, Hayes DL, et al. Long-term structural fail-
ure of coaxial polyurethane implantable cardioverter defibrillator
ac
leads. Pacing Clin Electrophysiol 2002; 25:879–882.
58. Hauser RG, Hayes DL, Almquist AK, et al. Unexpected ICD pulse
generator failure due to electronic circuit damage caused by elec-
trical overstress. Pacing Clin Electrophysiol 2001; 24:1046–1054.
59. Sweeney MO. Exploding implantable cardioverter defibrillator. J
Cardiovasc Electrophysiol 2001; 12:1422–1424.
60. Vollmann D, Luthje L, Vonhof S, et al. Inappropriate therapy and
.p
fatal proarrhythmia by an implantable cardioverter defibrillator.
Heart Rhythm 2005; 2:307–309.
61. Swerdlow CD, Cannom DS. ICD product reliability: Lessons from
electrical overstress failures. Pacing Clin Electrophysiol 2001;
24:1043–1045.
w
62. Wood M, Swerdlow C, Olson WH. Sensing and arrhythmia detec-
tion by implantable devices. In: Ellenbogen K, Kay GN, Wilkoff BL
(eds.): Clinical Cardiac Pacing and Defibrillation, 2nd Ed. Philade-
phia, W.B. Saunders, 2000, pp. 68–126.
63. Sweeney MO, Hellkamp AS, Ellenbogen KA, et al. Adverse effect
w
of ventricular pacing on heart failure and atrial fibrillation among
patients with normal baseline QRS duration in a clinical trial of
pacemaker therapy for sinus node dysfunction. Circulation 2003;
107:2932–2937.
w
64. Wilkoff BL, Cook JR, Epstein AE, et al. Dual-chamber pacing or ven-
tricular backup pacing in patients with an implantable defibrillator:
The Dual Chamber and VVI Implantable Defibrillator (DAVID) Trial
(see comment). JAMA 2002; 288:3115–3123.
65. Bristow MR, Saxon LA, Boehmer J, et al. Cardiac-resynchronization
therapy with or without an implantable defibrillator in advanced
chronic heart failure (see comment). N Engl J Med 2004; 350:2140–
2150.
66. Young JB, Abraham WT, Smith AL, et al. Combined cardiac resyn-
chronization and implantable cardioversion defibrillation in ad-
vanced chronic heart failure: The MIRACLE ICD Trial. JAMA 2003;
289:2685–2694.
67. Bradley DJ, Bradley EA, Baughman KL, et al. Cardiac resyn-
chronization and death from progressive heart failure: A meta-
analysis of randomized controlled trials (see comment). JAMA
2003; 289:730–740.
PACE, Vol. 29 January 2006
68. Asirvatham S, Hayes DL. Optimization of biventricular devices. In:
Hayes DL, Wang PJ, Sackner-Bernstein J, et al. (eds.): Resynchro-
nization and Defibrillation in Heart Failure: A Practical Approach.
Oxford, UK, Blackwell, 2004, pp. 139–162.
69. Lipchenca I, Garrigue S, Glikson M, et al. Inhibition of biventricular
pacemakers by oversensing of far-field atrial depolarization. Pacing
Clin Electrophysiol 2002; 25:365–367.
70. Garcia-Moran E, Mont L, Brugada J. Inappropriate tachycardia de-
tection by a biventricular implantable cardioverter defibrillator.
Pacing Clin Electrophysiol 2002; 25:123–124.
71. Bulava A, Ansalone G, Ricci R, et al. Triple-site pacing in patients
with biventricular device-incidence of the phenomenon and car-
diac resynchronization benefit. J Interv Card Electrophysiol 2004;
10:37–45.
72. Kay G. Troubleshooting and reprogramming of cardiac resynchro-
m
nization therapy. In: Ellenbogen KA, Kay GN, Wilkoff B (eds.): De-
vice Therapy for Congestive Heart Failure. Philadephia, Elsevier,
2004, pp. 294–310.
73. Knight BP, Desai A, Coman J, et al. Long-term retention of cardiac
resynchronization therapy. J Am Coll Cardiol 2004; 44:72–77.
co
74. Tse HF, Newman D, Ellenbogen KA, et al. Effects of ventricular rate
regularization pacing on quality of life and symptoms in patients
with atrial fibrillation (atrial fibrillation symptoms mediated by
pacing to mean rates [AF SYMPTOMS study]). Am J Cardiol 2004;
94:938–941.
75. Pacifico A, Hohnloser SH, Williams JH, et al. Prevention of
implantable-defibrillator shocks by treatment with sotalol. d,l-
Sotalol Implantable Cardioverter-Defibrillator Study Group. N Engl
J Med 1999; 340:1855–1862.
76. Rajawat YS, Dias D, Gerstenfeld EP, et al. Interactions of antiar-
rhythmic drugs and implantable devices in controlling ventric-
ular tachycardia and fibrillation. Curr Cardiol Rep 2002; 4:434–
440.
77. Moss AJ, Hall WJ, Cannom DS, et al. Improved survival with an
implanted defibrillator in patients with coronary disease at high
risk for ventricular arrhythmia. Multicenter automatic defibrilla-
tor implantation trial investigators. N Engl J Med 1996; 335:1933–
1940.
78. AVID Investigators. A comparison of antiarrhythmic-drug ther-
apy with implantable defibrillators in patients resuscitated from
near-fatal ventricular arrhythmias. N Engl J Med 1997; 337:1576–
1583.
79. Rajawat YS, Patel VV, Gerstenfeld EP, et al. Advantages and pitfalls
of combining device-based and pharmacologic therapies for the
treatment of ventricular arrhythmias. Pacing Clin Electrophysiol
2004; 27:1670–1681.
80. Hook BG, Perlman RL, Callans DJ, et al. Acute and chronic cycle
length dependent increase in ventricular pacing threshold. Pacing
Clin Electrophysiol 1992; 15:1437–1444.
81. Pelosi F Jr, Oral H, Kim MH, et al. Effect of chronic amiodarone
therapy on defibrillation energy requirements in humans. J Cardio-
vasc Electrophysiol 2000; 11:736–740.
82. Dohrmann ML, Goldschlager NF. Myocardial stimulation thresh-
old in patients with cardiac pacemakers: Effect of physiologic
variables, pharmacologic agents, and lead electrodes. Cardiol Clin
1985; 3:527–537.
83. Koul AK, Keller S, Clancy JF, et al. Hyperkalemia induced T wave
oversensing leading to loss of biventricular pacing and inappropri-
ate ICD shocks. Pacing Clin Electrophysiol 2004; 27:681–683.
84. Arthur W, Kaye GC. Hyperkalemia diagnosed by implantable car-
dioverter defibrillator T wave sensing. Pacing Clin Electrophysiol
2001; 24:908–909.
85. Moulik PK, Attar MN, Rose EL, et al. Successful resuscitation of a
patient with electrical storm. Emerg Med J 2003; 20:E4.
86. Pinski SL, Trohman RG. Interference in implanted cardiac devices,
part I. Pacing Clin Electrophysiol 2002; 25:1367–1381.
87. Pinski SL, Trohman RG. Interference in implanted cardiac devices,
part II. Pacing Clin Electrophysiol 2002; 25:1496–1509.
88. Groh WJ, Boschee SA, Engelstein ED, et al. Interactions be-
tween electronic article surveillance systems and implantable
cardioverter-defibrillators. Circulation 1999; 100:387–392.
89. McIvor ME, Reddinger J, Floden E, et al. Study of pacemaker and
implantable cardioverter defibrillator triggering by electronic arti-
cle surveillance devices (SPICED TEAS). Pacing Clin Electrophys-
iol 1998; 21:1847–1861.
90. Anfinsen OG, Berntsen RF, Aass H, et al. Implantable cardioverter
defibrillator dysfunction during and after magnetic resonance
imaging. Pacing Clin Electrophysiol 2002; 25:1400–1402.
91. Roguin A, Zviman MM, Meininger GR, et al. Modern pacemaker
95
 SWERDLOW AND FRIEDMAN
and implantable cardioverter/defibrillator systems can be magnetic
resonance imaging safe: In vitro and in vivo assessment of safety
and function at 1.5 T. Circulation 2004; 110:475–482.
92. Burke MC, Kopp DE, Alberts M, et al. Effect of radiofrequency cur-
rent on previously implanted pacemaker and defibrillator ventric-
ular lead systems. J Electrocardiol 2001; 34(Suppl.):143–148.
93. Casavant D, Haffajee C, Stevens S, et al. Aborted implantable car-
dioverter defibrillator shock during facial electrosurgery. Pacing
Clin Electrophysiol 1998; 21:1325–1326.
94. Medtronic Inc. Electromagnetic interference and the work site.
2003:3.
95. Wilkoff BL, Kuhlkamp V, Volosin K, et al. Critical analysis
of dual-chamber implantable cardioverter-defibrillator arrhythmia
detection: Results and technical considerations. Circulation 2001;
103:381–386.
96. Mitchell LB, Pineda EA, Titus JL, et al. Sudden death in patients
with implantable cardioverter defibrillators: The importance of
post-shock electromechanical dissociation. J Am Coll Cardiol 2002;
39:1323–1328.
97. Schaumann A, von zur Muhlen F, Herse B, et al. Empirical ver-
er
ac
.p
w
w
w
96 January 2006
sus tested antitachycardia pacing in implantable cardioverter de-
fibrillators: A prospective study including 200 patients. Circulation
1998; 97:66–74.
98. Wathen MS, Sweeney MO, DeGroot PJ, et al. Shock reduction using
antitachycardia pacing for spontaneous rapid ventricular tachy-
cardia in patients with coronary artery disease. Circulation 2001;
104:796–801.
99. Bardy GH, Lee KL, Mark DB, et al. Amiodarone or an implantable
cardioverter-defibrillator for congestive heart failure. N Engl J Med
2005; 352:225–237.
100. Ozahowski TP, Greenberg ML, Mock P, et al. Implantable car-
dioverter defibrillator clinic casualties: Inadvertent reprogramming
during routine implantable cardioverter defibrillator follow-up.
Pacing Clin Electrophysiol 1996; 19:1524–1525.
101. Swerdlow C. Optimal programming of sensing and detection in
m
single chamber ventricular ICDs. Card Electrophysiol Rev 2001;
5:85–90.
102. Gunderson BD, Patel AS, Bounds CA, et al. An algorithm to pre-
dict implantable cardioverter-defibrillator lead failure. J Am Coll
Cardiol 2004; 44:1898–1902.
co
d.
ic
PACE, Vol. 29